Philip Blundell writes:
> >Who, where, what initializes the region of memory referenced by
> >PARAMS_BASE.
>
> The bootloader is expected to do it. If you don't have a bootloader or it
> doesn't understand the Linux params system, you need to tell setup.c not to
> use the params structure.
Which would not be a good idea since then you'd have to have your own
patch that required to be applied to my kernels.
> >It appears that my start_mem and end_end are not initialized correctly.
> >It also appears that these variables are initialized via the
> >params_struct which appears to be uninitialized.
>
> That sounds very likely.
start_mem is set up by the kernel internally. It does not refer to the start
of memory, but to the start of free memory, typically located directly after
the running kernel. You therefore only have to worry about end_mem.
You want to set up page_size to be 4096 and nr_pages to be
(your memory size)/4096. Note, however, that the EBSA285 firmware does *not*
as far as I know (my firmware) detect the memory size and configure the 21285
to match the installed memory.
I do have a boot loader for Linux on the EBSA285 platform, but it's rather
limited at the moment - it supports 3com 3c59x, DEC Tulip (forgotten which
one(s)) and Promise UDMA33 (with UDMA configuration).
I'm still hoping to find some time (between Linux 2.2.0, work, sleep and hitting
Multitech over the head about their modems) to clean up the Promise UDMA driver
such that I can release the source code, but alas, something has always got in
the way over the last month.
_____
|_____| ------------------------------------------------- ---+---+-
| | Russell King [EMAIL PROTECTED] --- ---
| | | | http://www.arm.linux.org.uk/~rmk/armlinux.html / / |
| +-+-+ --- -+-
/ | THE developer of ARM Linux |+| /|\
/ | | | --- |
+-+-+ ------------------------------------------------- /\\\ |
unsubscribe: body of `unsubscribe linux-arm' to [EMAIL PROTECTED]
